CFSS TPS calculator

Open weights SenseTime,Chinese University of Hong Kong (CUHK),Shenzhen Institute of Advanced Technology 17.4K parameters June 2015

Each card below is assessed against this model at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from the card's memory bandwidth and the size of the model once compressed.

Calculated for this model

818 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Tesla C1080

4 GB · Q8_0 · 2,117,647 tok/s

Fastest card

B200

194,636,678 tok/s · 180 GB

Which GPUs can run CFSS?

Set the inputs, read the answer

Context length tokens

A longer conversation needs more memory, which can push this model off smaller cards.

Minimum quality

Hides cards that would only fit the model by compressing it below this point.

Search cards

Card memory

Also show cards that cannot run it

818 cards match

Calculating

					Needs	Quantisation	Fit
194,636,678 tok/s 116,782,007–311,418,685 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	0.7 GB	Q8_0	Comfortable
194,636,678 tok/s 116,782,007–311,418,685 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	0.7 GB	Q8_0	Comfortable
155,422,253 tok/s 93,253,352–248,675,606 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
155,422,253 tok/s 93,253,352–248,675,606 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
124,299,849 tok/s 74,579,909–198,879,758 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
118,971,670 tok/s 71,383,002–190,354,671 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
118,971,670 tok/s 71,383,002–190,354,671 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
113,862,457 tok/s 68,317,474–182,179,931 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	0.7 GB	Q8_0	Comfortable
101,052,930 tok/s 60,631,758–161,684,689 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
101,052,930 tok/s 60,631,758–161,684,689 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
101,052,930 tok/s 60,631,758–161,684,689 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
95,858,564 tok/s 57,515,138–153,373,702 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
81,747,405 tok/s 49,048,443–130,795,848 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
81,747,405 tok/s 49,048,443–130,795,848 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	0.7 GB	Q8_0	Comfortable
81,747,405 tok/s 49,048,443–130,795,848 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
81,747,405 tok/s 49,048,443–130,795,848 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
81,747,405 tok/s 49,048,443–130,795,848 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
62,244,810 tok/s 37,346,886–99,591,696 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
62,244,810 tok/s 37,346,886–99,591,696 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
51,870,675 tok/s 31,122,405–82,993,080 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
50,763,679 tok/s 30,458,207–81,221,886 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
49,632,353 tok/s 29,779,412–79,411,765 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	0.7 GB	Q8_0	Comfortable
49,632,353 tok/s 29,779,412–79,411,765 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	0.7 GB	Q8_0	Comfortable
49,632,353 tok/s 29,779,412–79,411,765 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	0.7 GB	Q8_0	Comfortable
49,632,353 tok/s 29,779,412–79,411,765 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable

Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.

On record

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation: SenseTime,Chinese University of Hong Kong (CUHK),Shenzhen Institute of Advanced Technology
Organisation type: Industry,Academia
Country: Hong Kong, China
Published: 7 June 2015
Authors: Shizhan Zhu, Cheng Li, Chen Change Loy, Xiaoou Tang

What it does

The problem areas the model was built for. A model can carry several of each.

Domain: Vision
Task: Face detection
Approach: Supervised

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters: 17.4K
Training data: 141,660 tokens

The training run

What it physically took to train: which chips, how many, for how long, and what that drew from the wall.

Chips used: 1

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights: Open — downloadable
Model access: Open weights (unrestricted)

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Why it is tracked: SOTA improvement
Record confidence: Likely

Sources

Where this record came from and when it was last checked.

Reference: Face alignment by coarse-to-fine shape searching
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run CFSS

Ranked by estimated tokens per second, newest card first where speeds tie. Because generation is bound by memory bandwidth, this ordering follows bandwidth rather than any gaming benchmark.

The smallest GPUs that still run CFSS

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

What the numbers mean

What you need to run it

Minimum card

Tesla C1080

Memory needed

0.7 GB

Fastest

194,636,678 tok/s

CFSS is small enough at 17.4K parameters that hardware is rarely the obstacle — 818 of the cards we track can run it, including cards several years old.

The least hardware that works is a Tesla C1080. Its 4 GB is enough at Q8_0 compression, giving roughly 2,117,647 tokens per second.

A B200 is the fastest we calculate for it: about 194,636,678 tokens per second, from 8,000 GB/s of memory bandwidth.

Background

CFSS was published by SenseTime,Chinese University of Hong Kong (CUHK),Shenzhen Institute of Advanced Technology, in Hong Kong, in June 2015. industry,Academia is the category the publisher falls under.

It works in Vision, and is recorded as doing face detection.

The weights are published, so it can be downloaded and run on your own hardware indefinitely, offline, with no account attached.

Reading the throughput figures

Half the cards that hold it manage more than 5,465,397.9 tokens per second, and 818 exceed reading speed outright.

It is a dense model, so every parameter is read for every token produced. That makes speed track memory bandwidth almost exactly — a card with twice the bandwidth generates roughly twice as fast.

Memory here is estimated from size rather than computed from the architecture, which is not recorded for this model — the numbers are indicative rather than exact.

How it was trained

It was trained on about 141,660 tokens of text.

Its inclusion criterion is sOTA improvement.

Step by step

How to choose a GPU for CFSS

The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.

01

Start from the memory column

The table lists every card that can hold CFSS — around 0.7 GB at Q8_0. That figure, not the card's headline performance, is what decides whether it runs.
02

Set the context length you will work at

Longer conversations cost memory on top of what the weights need. Move the slider to your real working length before trusting any row for CFSS.
03

Choose how far you will compress it

The quantisation column varies by card, because a bigger card holds a more accurate copy of CFSS — Q8_0 on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Compare tokens per second, not specifications

Ranking by tokens per second for CFSS follows memory bandwidth, not core counts, which is why the B200 tops it at 194,636,678 tok/s.
05

Look at the headroom, not just the fit

The fit column separates cards that just manage CFSS from those with room to spare. Buy for the second if the context might grow.
06

Open the card you have settled on

Following a card through to its own page shows every other model it can hold, which is the question that follows once CFSS is settled.

Answers

CFSS — common questions

Can I run CFSS on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 0.7 GB and generating roughly 32,601,644 tokens per second — a comfortable fit.

Is CFSS open source?

Its weights are published, so CFSS can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.

How many parameters does CFSS have?

CFSS has 17.4K parameters. The paper trains two regression models, each using hardcoded SIFT / BRIEF features. 68 keypoints (given in the paper) and 128 dimensional features per keypoint (assumption based on SIFT features) 2*68*128=17408. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

Who created CFSS?

CFSS was published by SenseTime,Chinese University of Hong Kong (CUHK),Shenzhen Institute of Advanced Technology, based in Hong Kong, categorised as industry,Academia.

When was CFSS released?

CFSS was published in June 2015. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

What is CFSS used for?

CFSS works in Vision, and is recorded as handling face detection. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download CFSS?

The weights for CFSS are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

Can I run CFSS if it does not fit in my GPU?

Partly. Layers that do not fit sit in system memory and run at a fraction of the speed, so a mostly-offloaded CFSS is rarely worth using. Every figure here assumes the whole model is on the card.

Would two GPUs run CFSS faster?

Capacity adds across cards; throughput does not. Since 818 of the cards we track already hold CFSS on their own, a second card is rarely the answer here.

Why does the quantisation differ between cards for CFSS?

Each card is shown running the least-compressed copy it can hold, and CFSS appears at 1 different compression levels across the cards that fit it. Bigger cards get the more accurate version.

How accurate are these CFSS speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 116,782,007–311,418,685 tok/s on the B200, for instance. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

What GPU do I need to run CFSS?

The smallest card in our catalogue that holds CFSS is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 0.7 GB, and produces roughly 2,117,647 tokens per second. 818 cards in total can run it.

How fast is CFSS on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 194,636,678 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 818 of the cards that can run CFSS clear that.

How much VRAM does CFSS need?

About 0.7 GB at Q8_0 compression, which is what the smallest card that runs it uses. Less compression needs more: the figures in the memory column above are recalculated for each card, because each one holds the least-compressed version it can.

Can I run CFSS on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 0.7 GB and generating roughly 36,251,081 tokens per second — a comfortable fit.

Can I run CFSS on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 0.7 GB and generating roughly 22,198,313 tokens per second — a comfortable fit.

Can I run CFSS on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 0.7 GB and generating roughly 27,492,431 tokens per second — a comfortable fit.

Source

Original publication

Record last updated 28 November 2025

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.

All AI models